• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• fall
• /
• pp.256-259
• /
• 2021

본 논문에서는 사람을 키넥트로 촬영하여 3 차원 아바타로 복원하여 연예인처럼 춤을 추게 하는 기술을 설계 구현하였다. 기존의 순수 딥러닝 기반 방식과 달리 본 기술은 3 차원 인체 모델을 사용하여 안정적이고 자유로운 결과를 얻을 수 있다. 우선 인체 모델의 기하학적 정보는 3 차원 조인트를 사용하여 추정하고 DensePose를 통하여 정교한 텍스쳐를 복원한다. 여기에 3 차원 포인트-클라우드와 ICP 매칭 기법을 사용하여 의상 모델 정보를 복원한다. 이렇게 확보한 신체 모델과 의상 모델을 사용한 아바타는 신체 모델의 rigged 특성을 그대로 유지함으로써 애니메이션에 적합하여 PSY 의 <강남스타일>과 같은 춤을 자연스럽게 표현하였다. 개선할 점으로 인체와 의류 부분의 좀 더 정확한 분할과 분할과정에서 발생할 수 있는 노이즈의 제거 등을 확인되었다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2021.06a
• /
• pp.154-156
• /
• 2021

최근 3차원 영상이 다양한 분야에서 활용되고 있으며, 이에 따라 3차원 영상에 대한 압축과 전송 방안에 대한 연구가 활발히 진행되고 있다. 국제 표준화 기구인 ISO/IEC 산하 Moving Picture Expert Group(MPEG)에서는 기존의 2차원 비디오 코덱을 이용하여 고밀도 포인트 클라우드 압축하는 방안인 V-PCC와 3DoF+ 영상을 압축하기 위한 방안인 MPEG Immersive Video(MIV)를 표준화 중에 있다. V-PCC와 MIV는 압축 방법의 유사성으로 인해 동일한 Volumetric Visual Video-based Coding(V3C) 형식으로 저장된다. 압축된 V3C 데이터를 효과적으로 저장하여 이용하기 위해서는 ISO based Media File Format(ISOBMFF) 캡슐화 과정이 필수적이다. 본 논문에서는 MPEG의 Carriage of V3C data 표준에 따라 V3C 데이터를 ISOBMFF로 캡슐화 실험을 진행하였으며, 실험에 대한 검증을 위하여 생성된 ISOBMFF 데이터를 V3C 데이터로 복원한 뒤, 디코딩 하여 확인하였다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.9
• /
• pp.1330-1339
• /
• 2022

3D object detection generally aims to detect relatively large data such as automobiles, buses, persons, furniture, etc, so it is vulnerable to small object detection. In addition, in an environment with limited resources such as embedded devices, it is difficult to apply the model because of the huge amount of computation. In this paper, the accuracy of small object detection was improved by focusing on local features using only one layer, and the inference speed was improved through the proposed knowledge distillation method from large pre-trained network to small network and adaptive quantization method according to the parameter size. The proposed model was evaluated using SUN RGB-D Val and self-made apple tree data set. Finally, it achieved the accuracy performance of 62.04% at mAP@0.25 and 47.1% at mAP@0.5, and the inference speed was 120.5 scenes per sec, showing a fast real-time processing speed.

• Journal of the Korea Computer Graphics Society
• /
• v.21 no.2
• /
• pp.1-10
• /
• 2015

In this paper, we propose an improved surface reconstruction method from an oriented point cloud. Our method is a classical least-square scheme, but is based on the 7-direction box-spline and the BCC (Body-Centered Cubic) lattice, which results in surfaces with superior quality and lower computational overhead, compared to other methods based on the B-splines on the Cartesian lattice. Specifically, when compared with two of the most popular techniques our method results in better surfaces but only takes ${\approx}53%$ computation time.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.1
• /
• pp.21-30
• /
• 2024

In this paper, we propose a method to build a sample dataset of the features of eight sensor-only facilities built as infrastructure for autonomous cooperative driving. The feature extracted from point cloud data acquired by LiDAR and build them into the sample dataset for recognizing the facilities. In order to build the dataset, eight sensor-only facilities with high-brightness reflector sheets and a sensor acquisition system were developed. To extract the features of facilities located within a certain measurement distance from the acquired point cloud data, a cylindrical projection method was applied to the extracted points after applying DBSCAN method for points and then a modified OTSU method for reflected intensity. Coordinates of 3D points, projected coordinates of 2D, and reflection intensity were set as the features of the facility, and the dataset was built along with labels. In order to check the effectiveness of the facility dataset built based on LiDAR data, a common CNN model was selected and tested after training, showing an accuracy of about 90% or more, confirming the possibility of facility recognition. Through continuous experiments, we will improve the feature extraction algorithm for building the proposed dataset and improve its performance, and develop a dedicated model for recognizing sensor-only facilities for autonomous cooperative driving.

• Journal of Digital Contents Society
• /
• v.16 no.5
• /
• pp.835-847
• /
• 2015

Photo scanning can create point cloud and polygon models like 3D scanners bringing an object into the 3 dimensional world by combining several sheets of photographic information. The created data give us information about planes and sectional forms required for a 2 dimensional survey as well as 3 dimensional figures of buried cultural property. It requires a lot of time to record buried cultural property in the field, however, the photo scanning technique does not need additional equipment and manpower so the work may begin immediately while the property is protected. Moreover, it reduces financial burdens as it creates 3 dimensional data using images acquired by photography but provides the optimal condition to check 3 dimensional information quickly and easily.

• Proceedings of the Korean Institute of Building Construction Conference
• /
• 2023.05a
• /
• pp.285-286
• /
• 2023

The visualization of information serves as a valuable tool for facilitating communication and exchange of opinions among stakeholders by conveying information in an intuitive and clear manner. As a preliminary study of visualization for construction field, this study proposed a model for generating three-dimensional floor layout using 360-degree panoramic cameras. The model integrates the layouts by calculating normal vectors of the plane which has openings, and applying translation and rotation matrices between the normal vectors. The results of this study can contribute to improving communication in construction sites by incorporating visualization, and further to the digital transformation of the construction industry.

• Korean Journal of Remote Sensing
• /
• v.37 no.5_3
• /
• pp.1425-1434
• /
• 2021

Continuous and periodic data acquisition must be preceded to maintain and manage the river facilities effectively. Adapting the existing general facilities methods, which include river surveying methods such as terrestrial laser scanners, total stations, and Global Navigation Satellite System (GNSS), has limitation in terms of its costs, manpower, and times to acquire spatial information since the river facilities are distributed across the wide and long area. On the other hand, the Mobile Mapping System (MMS) has comparative advantage in acquiring the data of river facilities since it constructs three-dimensional spatial information while moving. By using the MMS, 184,646,009 points could be attained for Anyang stream with a length of 4 kilometers only in 20 minutes. Levee points were divided at intervals of 10 meters so that about 378 levee cross sections were generated. In addition, the waterside maximum and average slope could be automatically calculated by separating slope plane form levee point cloud, and the accuracy of RMSE was confirmed by comparing with manually calculated slope. The reference slope was calculated manually by plotting point cloud of levee slope plane and selecting two points that use location information when calculating the slope. Also, as a result of comparing the water side slope with slope standard in basic river plan for Anyang stream, it is confirmed that inspecting the river facilities with the MMS point cloud is highly recommended than the existing river survey.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.20 no.3
• /
• pp.587-593
• /
• 2019

The field of geospatial information is rapidly changing due to the development of sensor and data processing technology that can acquire location information. And demand is increasing in various related industries and social activities. The construction and utilization of three dimensional geospatial information that is easy to understand and easy to understand can be an essential element to improve the quality and reliability of related services. In recent years, 3D laser scanners are widely used as 3D geospatial information construction technology. However, 3D laser scanners may cause shadow areas where data acquisition is not possible when objects are large in size or complex in shape. In this study, 3D model of an object has been created by acquiring oblique images using an unmanned aerial vehicle and processing the data. The study area was selected, oblique images were acquired using an unmanned aerial vehicle, and point cloud type 3D model with 0.02 m spacing was created through data processing. The accuracy of the 3D model was 0.19m and the average was 0.11m. In the future, if accuracy is evaluated according to shooting and data processing methods, and 3D model construction and accuracy evaluation and analysis according to camera types are performed, the accuracy of the 3D model will be improved. In the point cloud type 3D model, Cross section generation, drawing of objects, and so on, it is possible to improve work efficiency of spatial information service and related work.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2018.05a
• /
• pp.465-468
• /
• 2018

현재 사용되고 있는 보편적인 이미지 편집 방식은 이미지 내부 일부 영역을 선택 및 추출하는 방식으로 객체를 배경과 분리한다. 객체가 분리되는 과정에서 객체가 있었던 곳에서는 빈 영역이 발생하게 되는데, 이 문제를 해결하기 위해 인접한 영역을 가져와서 채우거나, 딥러닝을 적용하여 유사한 이미지로 채우는 방식이 가장 보편적이다. 그러나 이러한 방식은 배경에서 유실된 부분을 인공적인 방법으로 채우기 때문에 완벽하게 복원하기가 힘들다. 따라서 본 논문에서는 미리 해당 이미지에 대한 3 차원 정보를 가공 및 저장함으로써 편집으로 인해 유실되는 부분을 3 차원 정보로 부터 복구할 수 있는 아이디어를 제안한다.